U.S. patent application number 17/690958 was filed with the patent office on 2022-09-22 for detecting and quantifying a liquid and/or food intake of a user wearing a hearing device.
The applicant listed for this patent is SONOVA AG. Invention is credited to Manuela Feilner.
Application Number | 20220301683 17/690958 |
Document ID | / |
Family ID | 1000006253766 |
Filed Date | 2022-09-22 |
United States Patent
Application |
20220301683 |
Kind Code |
A1 |
Feilner; Manuela |
September 22, 2022 |
DETECTING AND QUANTIFYING A LIQUID AND/OR FOOD INTAKE OF A USER
WEARING A HEARING DEVICE
Abstract
A method for detecting and quantifying a liquid and/or food
and/or medication intake of a user wearing a hearing device which
comprises at least one microphone. The method comprises: receiving
an audio signal from the at least one microphone and/or a sensor
signal from at least one further sensor; and collecting and
analyzing the received audio signal and/or further sensor signals
so as to detect each time the user drinks and/or takes medication
and/or eats something, wherein drinking and/or medication intake is
distinguished from eating and/or wherein drinking is distinguished
from medication intake, and so as to determine values indicative of
how often this is detected and/or a respective amount of liquid
and/or food and/or medication ingested by the user.
Inventors: |
Feilner; Manuela; (Egg b.
Zurich, CH) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
SONOVA AG |
Staefa |
|
CH |
|
|
Family ID: |
1000006253766 |
Appl. No.: |
17/690958 |
Filed: |
March 9, 2022 |
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
G16H 50/30 20180101;
G16H 20/10 20180101; G16H 20/60 20180101; G06N 3/04 20130101; H04R
1/08 20130101; H04R 1/1016 20130101 |
International
Class: |
G16H 20/60 20060101
G16H020/60; G16H 50/30 20060101 G16H050/30; G16H 20/10 20060101
G16H020/10; H04R 1/10 20060101 H04R001/10; H04R 1/08 20060101
H04R001/08; G06N 3/04 20060101 G06N003/04 |
Foreign Application Data
Date |
Code |
Application Number |
Mar 22, 2021 |
EP |
EP21163914 |
Claims
1. A method for detecting and quantifying a liquid and/or food
and/or medication intake of a user wearing a hearing device which
comprises at least one microphone, the method comprising: receiving
an audio signal from the at least one microphone and/or a sensor
signal from at least one further sensor; collecting and analyzing
the received audio signal and/or further sensor signals so as to
detect each time the user drinks and/or takes medication and/or
eats something, wherein drinking and/or medication intake is
distinguished from eating and/or wherein drinking is distinguished
from medication intake, and so as to determine values indicative of
how often this is detected and/or a respective amount of liquid
and/or food and/or medication ingested by the user; wherein the
step of analyzing includes applying one or more machine learning
algorithms in the hearing device or in a hearing system, part of
which the hearing device is, or in a remote server or cloud
connected to it; and storing the determined values in the hearing
system and, based on the stored values, generating a predetermined
type of output.
2. The method of claim 1, wherein, in the step of analyzing, at
least one of the machine learning algorithms is applied in its
training phase so as to learn user-specific manners of drinking
and/or eating and/or medication intake; and the newly learned
user-specific manners are incorporated in the future analysis
step.
3. The method of claim 1, wherein, in the step of analyzing, two or
more different phases of drinking or, respectively, eating or,
respectively, medication intake, are distinguished in the course of
detecting a liquid and/or food and/or medication intake of the
user; and the analysis of the different phases is based on signals
from correspondingly different sensors and/or is performed by
correspondingly different machine learning algorithms.
4. The method of claim 3, wherein the different phases of drinking
and/or medication intake comprise one or more of the following
phases: bringing a source of liquid in contact with the mouth,
based at least on a signal from at least one movement sensor and/or
orientation sensor sensing a corresponding movement of some upper
body part of the user; tilting of the user's head, based at least
on a signal from at least one movement sensor sensing a
corresponding movement of the head of the user and/or based at
least on a signal from at least one orientation sensor sensing a
corresponding orientation of the head of the user relative to the
surface of the earth; gulping or sipping the liquid and/or
swallowing the medication, based at least on a signal from the at
least one microphone and/or on a signal from at least one movement
sensor sensing a corresponding movement of the user's throat, head
and/or breast; removing the mouth from the source of liquid, based
at least on a signal from the at least one microphone and/or on a
signal from at least one movement sensor sensing a corresponding
movement of some upper body part of the user.
5. The method of claim 4, wherein the different phases of
medication intake further comprise one or more of the following
phases: bringing, before the source of liquid is brought in contact
with the mouth, a medication in contact with the mouth and/or
inserting the medication into the mouth, based at least on a signal
from at least one movement sensor and/or orientation sensor sensing
a corresponding movement of some upper body part of the user.
6. The method of claim 4, wherein drinking is distinguished from
medication intake by a different tilting angle of the user's head
relative to the surface of the earth, based at least on the signal
from the at least one movement sensor and/or the at least one
orientation sensor.
7. The method of claim 1, wherein the further sensor signals
comprise physiological signals indicative of a physiological
property of the user collected by at least one physiological
sensor, and wherein, in the step of analyzing, an event of drinking
or, respectively, eating or, respectively, medication intake and/or
which kind of liquid or, respectively, food or, respectively,
medication the user is taking is further determined based on the
physiological property.
8. The method of claim 7, wherein the physiological signals are
indicative of at least one of a cardiovascular property, a body
fluid analyte level, and a body temperature.
9. The method of claim 7, wherein an event of water intake and/or
an amount of water ingested by the user during the drinking or,
respectively, eating or, respectively, medication intake is
estimated based on the physiological property.
10. The method of claim 1, wherein at least one of the machine
learning algorithms is based on an artificial neural network; the
input data set for the neural network is provided at a respective
time point by the sensor data collected over a predetermined period
of time up to this time point; the output data set for the
respective time point includes a frequency or number of detected
liquid and/or food and/or medication intakes as well as a
respective or an overall amount of the liquid and/or food and/or
medication ingested by the user and/or a duration of the detected
liquid and/or food and/or medication intakes; wherein the learning
phase is implemented by a supervised learning, in which the
algorithm is trained using a database of input sensor data with
labeled output data sets; or, alternatively, by an unsupervised
learning in an environment with more information available and/or
by a reinforcement learning or deep reinforcement learning.
11. The method of claim 10, wherein the artificial neural network
is a deep neural network including at least one hidden layer.
12. The method of claim 1, wherein, in the step of analyzing, a
temporal dynamic behavior of the drinking and/or eating and/or
medication intake process, is incorporated by applying at least one
of the following machine learning methods: a Hidden Markow Model; a
recurrent neural network.
13. The method of claim 1, wherein, in the step of analyzing, a
dehydration risk of the hearing device user is estimated depending
on the determined values of the amount and of a frequency of the
user's liquid intake; and the generated output is configured
depending on the estimated dehydration risk, so as to counsel the
user to ingest a lacking amount of liquid and/or so as to inform
the user and/or a person close to the user and/or a health care
professional about the estimated dehydration risk.
14. The method of claim 1, wherein an interactive user interface is
provided in the hearing system; and the steps of analyzing and/or
generating an output are supplemented by an interaction with the
user via the interactive user interface, wherein the user is
enabled to input additional information pertaining to his liquid
and/or food and/or medication intake.
15. The method of claim 14, wherein additional information about a
need to take a predetermined medication is stored in the hearing
system; when a fluid intake of the user is detected, the output is
generated depending on this additional information and comprises
questioning the user, via the interactive user interface, whether
he has taken the predetermined medication; and the user's response
to this question via the interactive user interface is stored in
the hearing system and/or transmitted to the user and/or a person
close to the user and/or a health care professional, so as to
verify that the user has taken the predetermined medication.
16. The method of claim 1, wherein, in the step of generating an
output, depending on the determined frequency and amount of the
liquid or, respectively, food or, respectively, medication ingested
by the user, an output configured such as to enhance the user's
desire to drink or, respectively, to eat something, or,
respectively, to take the medication is generated by augmented
reality means in the hearing system.
17. The method of one of claim 1, wherein when detecting that the
user is drinking or, respectively, eating something or,
respectively, taking medication and/or upon detecting which kind of
liquid or, respectively, food or, respectively, medication the user
is taking, an output configured such as to enhance the user's
experience of drinking or, respectively, eating or, respectively,
taking medication is generated by augmented reality means in the
hearing system depending on this detection.
18. A computer-readable medium, in which a computer program is
stored for detecting and quantifying a liquid and/or food and/or
medication intake of a user wearing a hearing device which
comprises at least one microphone, which program, when being
executed by a processor, is adapted to carry out the steps of the
method of claim 1.
19. A hearing device worn by a hearing device user, comprising: a
microphone; a processor for processing a signal from the
microphone; a sound output device for outputting the processed
signal to an ear of the hearing device user; wherein the hearing
device is adapted for performing the method of claim 1.
Description
RELATED APPLICATIONS
[0001] The present application claims priority to EP Patent
Application No. EP21163914, filed Mar. 22, 2021, the contents of
which are hereby incorporated by reference in their entirety.
BACKGROUND INFORMATION
[0002] Hearing devices are generally small and complex devices.
Hearing devices can include a processor, microphone, an integrated
loudspeaker as a sound output device, memory, housing, and other
electronical and mechanical components. Some example hearing
devices are Behind-The-Ear (BTE), Receiver-In-Canal (RIC),
In-The-Ear (ITE), Completely-In-Canal (CIC), and
Invisible-In-The-Canal (IIC) devices. A user can prefer one of
these hearing devices compared to another device based on hearing
loss, aesthetic preferences, lifestyle needs, and budget.
[0003] Many of the elderly often do not drink sufficiently, because
of several reasons, e. g. lack of thirst, lack of remembering that
they shall drink. Subsequently, they have a high risk to dehydrate,
which might cause problems such as dizziness, a condition of mental
decline etc.
[0004] Hearing devices can be used to monitor the drinking behavior
of their users and to counsel the users by helping them to develop
a healthy hydrated lifestyle. To this end, the hearing system shall
detect automatically the fluid intake of a user.
[0005] In this context, for example, DE 10 2018 204 695 A1
discloses using hearing systems, which comprise hearing devices or
other devices worn by the user on his head, such as earphones or
headsets, for health monitoring. The document proposes recognizing
a large variety of specific symptoms or irregularities in the
breathing, speaking, sleeping, walking, chewing, swallowing or
other body functions of the user by means of sensors such as
microphones, vibration sensors or accelerometers e. g. integrated
in the hearing device worn by the user. Use cases address symptoms
of various diseases such as Parkinson, traumatic brain injury,
tics, bruxism, hiccup, allergic reactions, coughing fits and many
more. Inter alia, swallowing is proposed to be recognized by
sensing a typical sound moving downwards away from the hearing
device with a microphone, in combination with multiple
characteristic muscle contractions sensed by a vibration sensor in
the ear canal. It is also mentioned that a single gulp may be
interpreted as swallowing saliva, but possibly also as medication
intake, which may be used to monitor the latter. It is also briefly
mentioned that a drinking process might be detected as more or less
regular gulps by a microphone or a vibration sensor, and that the
user may also be reminded to drink. However, no detailed evaluation
algorithms are described for these different use cases.
[0006] A specific method of detecting a dehydration by measuring a
water level in the brain of the hearing device users, or a bio
impedance, is disclosed in US 2017/0289704 A1, based on the
hypothesis that the magnetic/electric conductance in the head
varies with the relative water level in the head, at least on a
short term. If the measured water level is below a predefined
threshold, a reminder signal reminding the user to drink something
is generated by at least one of the hearing devices and sent to the
user via e.g. a smartphone or a smartwatch or as a direct audio
reminder, e.g. a speech message.
[0007] Another problem known in the art is that the elderly often
lose their appetite, whereas bad eating habits may increase obesity
or the risk of heart diseases.
[0008] A further problem known in the art is that the elderly often
forget to take their prescribed medication or intentionally abstain
from taking their medication by underestimating an underlying
health risk.
BRIEF DESCRIPTION OF THE DRAWINGS
[0009] Below, embodiments of the present invention are described in
more detail with reference to the attached drawings.
[0010] FIG. 1 schematically shows a hearing system according to an
embodiment.
[0011] FIG. 2 shows a flow diagram of a method according to an
embodiment for detecting and quantifying a liquid and/or food
and/or medication intake of a user wearing a hearing device of the
hearing system of FIG. 1.
[0012] The reference symbols used in the drawings, and their
meanings, are listed in summary form in the list of reference
symbols.
DETAILED DESCRIPTION
[0013] Described herein are a method, a computer program and a
computer-readable medium for detecting and quantifying a liquid
and/or food and/or medication intake of a user wearing a hearing
device which comprises at least one microphone. Furthermore, the
embodiments described herein relate to a hearing system which
comprises at least one hearing device of this kind and optionally a
connected user device, such as a smartphone and/or a
smartwatch.
[0014] It is a feature described herein to provide a reliable and
robust method and system for detecting and/or quantifying a liquid
and/or food and/or medication intake of a user wearing a hearing
device. It is a further feature to provide a reliable and robust
method of detecting dehydration and/or monitoring a correct intake
of medication. It is a further feature to enhance the user's
experience of drinking or, respectively, eating, or, respectively,
medication intake by augmented reality means.
[0015] A first aspect relates to a method for detecting and
quantifying a liquid and/or food and/or medication intake of a user
wearing a hearing device which comprises at least one
microphone.
[0016] The method may be a computer-implemented method, which may
be performed automatically in the hearing device and/or in another
device of a hearing system, part of which the user's hearing device
is. As described in more detail herein below, the hearing system
may consist of the hearing device alone, or may be a binaural
hearing system comprising two hearing devices worn by the same
user, or may comprise the hearing device and a remote device
portable by the user, such as a smartphone or smartwatch, connected
to the hearing device. One or both of the hearing devices may be
worn on and/or in an ear of the user. A hearing device may be a
hearing aid, which may be adapted for compensating a hearing loss
of the user. Also a cochlear implant may be a hearing device.
[0017] According to an embodiment, the method comprises receiving
an audio signal from the at least one microphone and/or a sensor
signal from at least one further sensor. As described in more
detail herein below, the further sensors may include any types of
physical or physiological sensors--e.g. movement sensors, such as
accelerometers, and/or optical sensors, such as cameras, and/or
physiological sensors such as (body) temperature sensors and/or
heart rate sensors and/or photoplethysmography (PPG) sensors and/or
bioelectric sensors (e.g., electrocardiography (ECG) sensors,
electroencephalogram (EEG) sensors, electrooculography (EOG)
sensors, etc.) and/or blood analyte sensors (e.g., optical sensors
or radio frequency (RF) sensors sensitive to specific frequencies
of an analyte in the blood such as, e.g., glucose, water,
hemoglobin, etc., and/or voltametric sensors configured for
voltametric measurements indicating a presence of an electroactive
substance, e.g., a drug or a drug component, contained in a body
fluid, e.g. sweat.), etc.--integrated in the hearing device or
possibly also in a connected user device, such as a smartphone or a
smartwatch.
[0018] According to an embodiment, the method further comprises
collecting and analyzing the received audio signal and/or further
sensor signals so as to detect each time the user drinks and/or
takes medication and/or eats something, wherein drinking and/or
medication intake is distinguished from eating and/or wherein
drinking is distinguished from medication intake. For example,
drinking may be distinguished from eating and/or medication intake
may be distinguished from eating and/or drinking may be
distinguished from medication intake. As another example, drinking
(irrespective whether the drinking may be accompanied by a
medication intake or whether the drinking occurs without a
medication intake) may be distinguished from eating. As another
example, a medication intake (irrespective whether the medication
intake may be accompanied by drinking or whether the medication
intake occurs without drinking) may be distinguished from eating.
As a further example, drinking may be distinguished from a
medication intake, e.g., drinking without medication intake may be
distinguished from drinking with medication intake. As yet another
example, drinking may be distinguished from a medication intake and
drinking may also be distinguished from eating and medication
intake may also be distinguished from eating. The analysis further
includes determining values indicative of how often this is
detected and/or a respective amount of liquid and/or food and/or
medication ingested by the user.
[0019] The determined values may then be stored in the hearing
system and, based on the stored values, a predetermined type of
output is generated, in particular, to the user and/or to a third
party, such as a person close to the user and/or a health care
professional. The predetermined type of output may, for example,
include one or more of the following: a textual, acoustical,
graphical, vibrational output, as well as an output generated using
augmented reality and/or an interactive user interface etc. The
specific choice of a suitable output may, for instance, depend on
the specific goals and use cases addressed in the different
embodiments as described herein below.
[0020] According to an embodiment, the above-mentioned step of
analyzing includes applying one or more machine learning
algorithms, subsequently and/or in parallel, in the hearing device
or in the hearing system or in a remote server or cloud connected
to the hearing device or system. By applying a suitable machine
learning algorithm, a most reliable and/or robust method of
detecting and quantifying the liquid and/or food intake of the user
may be provided.
[0021] This may, for example, be achieved due to the capability of
a machine learning algorithm to automatically adapt the analysis to
individual drinking and/or eating habits and/or medication intake
habits of the user, which may also take into account that habits of
one and the same user are often changing with time, e.g. vary from
season to season or may even rapidly change with a changing life
style when the user moves to another country, changes his job,
marries etc. Furthermore, the reliability and robustness may also
be increased by a machine learning algorithm which is configured to
automatically supplement the analysis by adding new sensor-based
recognition criteria of the user's drinking and/or eating and/or
medication intake behavior, after having started with some basic or
initial recognition criteria. In the course of this, for example,
sensor signals of additional sensors provided in the hearing system
may be automatically included in the analysis algorithm by machine
learning.
[0022] Therefore, according to an embodiment, at least one of the
machine learning algorithms may be applied in its training phase so
as to learn user-specific manners of drinking and/or eating and/or
taking medication, in order to reliably detect whenever the user
drinks or eats something or takes medication and to distinguish
eating from drinking and/or medication intake and/or to distinguish
drinking from medication intake, e.g., to distinguish eating from
drinking and/or to distinguish eating from medication intake and/or
to distinguish drinking from medication intake. For example, the
machine learning algorithms may be applied to learn that--or how
exactly--the user typically elevates his head slightly to take a
gulp from a bottle or a sip from a glass or cup of liquid; or to
learn that--or how exactly--the user is typically lowering his head
when taking a bite when eating; or to learn that the user is
typically raising and/or tilting his head even more when swallowing
a medication in conjunction with drinking as compared to during
regular drinking without a medication intake. The newly learned
user-specific manners are then incorporated in the future analysis
step. Further details and examples are described herein below.
[0023] Any method features described herein with respect to a
drinking process (liquid intake) may also be applied by analogy to
detecting and quantifying an eating process (food intake) of the
hearing device user and/or by analogy to detecting and quantifying
a medication intake process (e.g., swallowing a pill or medical
syrup which may be in conjunction with a drinking process such as
drinking water, or in the absence of such a drinking process) of
the hearing device user. The detection algorithm is then
correspondingly adapted to use and interpret the collected sensor
signals so as to distinguish eating from drinking (e.g., drinking
with or without an accompanying medication intake), and/or drinking
is distinguished from medication intake and/or eating is
distinguished from medication intake, and so as to learn the
respective specific recognition and quantification criteria.
[0024] Particularly important examples of respective output
generation in either case are described in more detail herein
below. They provide, for example, novel and/or improved methods of
dehydration detection, verification of a medication intake,
enhancement of a liquid intake desire, more particularly a water
intake desire as compared to more unhealthy drinking habits, and/or
enhancement of an eating or drinking or medication intake
experience of the hearing device user, so as to raise the user's
life quality and support a healthy living style.
[0025] According to an embodiment, in the step of analyzing, two or
more different (in particular, subsequent) phases of drinking or,
respectively, eating or, respectively, medication intake, are
distinguished in the course of detecting a liquid and/or food
and/or medication intake of the user. Thereby, again, the
robustness and reliability of the method may be increased.
[0026] Since different sensors might be better suited to detect the
different phases, the analysis of the different phases may be based
on signals from at least partly different sensors or groups of
sensors. On the other hand, the analysis of different phases may
also be performed by at least partly different machine learning
algorithms. In the case of subsequent phases, the respective
algorithms may also be applied in a corresponding subsequent
manner.
[0027] In this embodiment, the different phases of drinking may,
for example, comprise one or more of the following phases (whereas
the different phases of eating may be defined in a similar manner
by analogy):
[0028] A phase of drinking (e.g., drinking with or without an
accompanying medication intake) and/or medication intake may be
bringing a source of liquid in contact with the mouth. This may,
for example, be grabbing a glass or bottle or another vessel
containing the liquid. Alternatively, this may also be lowering the
head so as to reach a jet of water, e. g. emanating from a water
tap, with the mouth. Detecting this phase may be based, inter alia,
on a signal from at least one movement sensor sensing a
corresponding movement of some upper body part of the user and/or
an orientation sensor sensing a corresponding change of orientation
of the body part relative to the surface of the earth. Suitable
movement sensors may include movement sensors, e.g. accelerometers,
provided at the user's head, e.g. in the hearing device, and/or at
the user's wrist or finger, e.g. in a wrist band, smartwatch or
finger ring, and/or at another upper body part of the user. An
accelerometer may also be employed as an orientation sensor, e.g.
with respect to determining an orientation relative to the
direction of the gravitational force acting perpendicularly to the
earth's surface.
[0029] Another phase of drinking and/or medication intake may be
tilting of the user's head. E.g., the user may tilt his head from a
more downward (e.g., toward the earth's surface) directed direction
(e.g., when eating, or when performing another activity such as
reading) to a more upward directed direction (e.g., to facilitate
the act of swallowing the liquid and/or medication). Such a user
behavior may be detected based at least on a signal from at least
one movement sensor sensing a corresponding movement of the head of
the user and/or based at least on a signal from at least one
orientation sensor sensing a corresponding orientation of the head
of the user relative to the surface of the earth, e.g. an
accelerometer.
[0030] Another phase of drinking and/or medication intake may be
gulping or sipping the liquid and/or swallowing the medication.
Detecting this phase may be based, inter alia, on a signal from the
at least one microphone, e.g. in a hearing device. In some
instances, determining a number of sounds related to the drinking
and/or medication intake, e.g. a number of temporally separated
gulping, sipping, and/or swallowing sounds, can be employed to
quantify the fluid intake and/or medication intake. Detecting this
phase may also be based, e.g., on a signal from at least one
movement sensor sensing a corresponding movement of the user's
throat, head and/or breast. Suitable movement sensors may include
at least one accelerometer provided in the hearing device and/or on
a head, neck or breast of the user. Quantifying the fluid intake
and/or medication intake may also be based on determining a number
of those movements.
[0031] Another phase of drinking and/or medication intake may be
removing the mouth from the source of liquid. This may be, for
example, putting the glass, bottle or another vessel back on the
table. Alternatively, this may also be raising the head away from
the jet of water. Detecting this phase may be based, inter alia, on
a signal from the at least one microphone and/or on a signal from
at least one movement sensor sensing a corresponding movement of
some upper body part of the user. Suitable movement sensors may
include movement sensors and/or orientation sensors, such as
accelerometers, provided at the user's head, e.g. in the hearing
device, and/or at the user's wrist or finger, e.g. in a wrist band,
smart watch or finger ring, and/or at another upper body part of
the user.
[0032] Another phase of medication intake may be bringing a
medication in contact with the mouth and/or inserting the
medication into the mouth, e.g., before a source of liquid is
brought in contact with the mouth. Detecting this phase may also be
based, e.g., on a signal from at least one movement sensor and/or
orientation sensor sensing, e.g. at least one accelerometer in the
hearing device and/or at the user's wrist or finger and/or at
another upper body part of the user. E.g., the medication may be a
pill, a syrup, a droplet and/or the like. E.g., a source of liquid
may be subsequently brought in contact with the mouth and swallowed
to facilitate a swallowing of the medication. In some instances,
medication intake may be distinguished from drinking, e.g., from a
regular drinking activity not involving a medication intake, by
detecting this phase.
[0033] In some instances, drinking, e.g., a drinking activity not
involving a medication intake, may be distinguished from a
medication intake, e.g., a drinking activity involving a medication
intake, by a different tilting angle of the user's head relative to
the surface of the earth. To illustrate, when swallowing a
medication, the user may tilt his head to the back. E.g. the
tilting angle may be larger during taking of medication as compared
to a tilting angle occurring during drinking. The tilting angle of
the head may be detected, e.g., based at least on a signal from at
least one movement sensor and/or at least one orientation
sensor.
[0034] Another phase of drinking and/or medication intake and/or
eating may be an impact on a physiological property of the user
caused by the drinking and/or medication intake and/or eating.
E.g., drinking liquid or eating nutrition may alter a
cardiovascular property of the user, e.g. a heart rate, and/or
change a blood analyte level, e.g., an amount of glucose and/or
lipid and/or water contained in the blood, and/or can also have an
impact on the body temperature (e.g., after consuming cold or hot
liquid or food and/or after consuming a large amount of food or
liquid). As another example, a medication intake can also alter a
cardiovascular property of the user, e.g. a blood pressure and/or
heart rate, and/or can also change a blood analyte level, e.g., a
drug or drug component of the medication contained in the blood. A
concentration of the blood analyte level, e.g., glucose, water,
lipid, a drug component, etc., may also be determined, e.g. to
quantify the ingested liquid and/or medication and/or food.
Detecting this phase may be based on further sensor signals, which
may comprise physiological signals indicative of a physiological
property of the user, which may be collected by at least one
physiological sensor. E.g., the physiological sensor(s) may be
included in the hearing device and/or they may be worn at another
body portion of the user than the ear, for instance at a wrist
(e.g., included in a smartwatch) or a finger (e.g., included in a
finger ring) and/or any other upper or lower body portion suitable
to detect a physiological property of the user. Correspondingly, an
event of drinking or, respectively, eating or, respectively,
medication intake and/or which kind of liquid or, respectively,
food or, respectively, medication the user is taking may be further
determined based on the physiological property.
[0035] A physiological property, as determined by a physiological
sensor in a physiological signal, may comprise any measurable
biological characteristic of a human being, e.g., the user, such as
a vital sign and/or a physiological property of the human being.
The physiological property may be measured by detecting any form of
energy and/or matter intrinsic to the human being and/or emitted
from the human being and/or caused by the human being. In some
implementations, the physiological signals are indicative of at
least one of a cardiovascular property (e.g., a heart rate and/or a
blood pressure and/or a blood oxygen saturation level, etc.), a
body fluid analyte level (e.g., a concentration of an analyte, such
as hemoglobin, lipid, glucose, water, a drug component, etc., in a
body fluid, e.g., in blood and/or in sweat, etc.), and a body
temperature.
[0036] In some implementations, a physiological sensor configured
to provide physiological signals indicative of a physiological
property comprises a light source configured to emit light through
a skin of the user and an optical detector for detecting a
reflected and/or scattered part of the light, wherein the
physiological signals comprise information about the detected
light. In particular, the physiological signals may comprise
information about a blood flow, e.g., blood volume changes, a heart
rate, a blood pressure, a blood oxygen saturation level, etc.,
indicated in an photoplethysmography (PPG) measurement. The
physiological signals may also comprise information about a blood
analyte level, e.g. by detecting an absorption and/or emission
spectrum of specific molecules contained in the blood, e.g., water
and/or lipids and/or glucose and/or a drug component. In some
implementations, the physiological sensor comprises an electrode
configured to detect an electric signal induced through a skin of
the user, wherein the physiological signals comprise information
about the electric signal. In particular, the physiological signals
may comprise information about a brain activity indicated in an
electroencephalogram (EEG) measurement and/or information about a
heart activity indicated in an electrocardiogram (ECG) measurement
and/or information about an eye activity indicated in an
electrooculography (EOG) measurement. In some implementations, the
physiological sensor comprises a temperature sensor configured to
detect a body temperature of the user, wherein the physiological
signals comprise information about the body temperature. In some
implementations, the physiological sensor comprises a radio
frequency (RF) sensor configured to send energy at a radio
frequency into tissue of the user and to detect a reflection and/or
absorption thereof, for instance to determine a blood analyte
level, e.g. an amount and/or density of certain molecules. In some
implementations, the physiological sensor comprises a voltametric
sensor configured to detect a voltametric property indicating a
presence of an analyte as an electroactive substance, e.g., a drug
or a drug component, which may be contained in a body fluid, e.g.
sweat.
[0037] Detecting the physiological property may also be employed in
the detection of liquid intake to determine a type of the ingested
liquid, e.g., to distinguish an event, in which the user is
drinking (pure) water, from another event, in which the user is
drinking a liquid different from (pure) water or containing less
water. To illustrate, the user may have rather unhealthy drinking
habits and may drink not enough water but rather prefers other
beverages which are dehydrating (e.g., coffee, alcoholic beverages,
etc.) or which contain a considerable amount of sugar. E.g.,
detecting a blood analyte level such as glucose and/or water can
give a direct indication whether a drinking event is related to
water consumption or another type of liquid and/or detecting a
cardiovascular property can give an indirect indication thereof
(e.g., consuming coffee can increase the blood pressure).
Determining the type of the ingested liquid, e.g., water as
compared to a different liquid type, can be applied in a
dehydration detection as described herein above and below.
[0038] Correspondingly, the detection algorithm might also consist
of a plurality of parts, e.g. two or three or more parts, which are
processed in a timely sequence, according to the above described
two or three or more drinking phases. Furthermore, instead of the
above-mentioned exemplary two or three or more different phases,
another number and/or types of different phases may be identifiable
in the course of detecting a drinking or eating or medication
intake process of the user, for example one or two or three or four
or more subsequent phases.
[0039] According to an embodiment, at least one of the machine
learning algorithms applied in the analysis step is based on an
artificial neural network. The input data set for the neural
network may be provided at a respective time point by the sensor
data collected over a predetermined period of time up to this time
point. The output data set for the respective time point may
include a frequency or number of detected liquid and/or food and/or
medication intakes and/or a respective or an overall amount of the
liquid and/or food and/or medication ingested by the user and/or a
duration of the detected liquid and/or food and/or medication
intakes. The artificial neural network may, for example, be a deep
neural network including one or multiple hidden layers.
[0040] The training phase of the (deep) neural network may be
implemented by a supervised learning, in which the algorithm is
trained using a database of labeled input sensor data with
corresponding output data sets. A suitable database of training
data, working well for the above-described analysis, contains, for
example, recorded swallowing or, respectively, gulping sounds of a
large number of people--in this example approximately 1000 adults.
Alternatively or additionally, representative sound recordings of
drinking, swallowing, gulping, chewing people available in the
Internet may also be used as training data.
[0041] Furthermore, the training phase may be implemented by an
unsupervised learning in an environment with still more information
available (Internet of Things), e.g. by using data transmitted from
smart devices related to drinking and eating and medication intake,
such as smart water bottles known in the art, and/or additional
information within a smart home and/or input from the user via an
interactive user interface as described herein below. The training
phase may also comprise reinforcement learning or deep
reinforcement learning.
[0042] Moreover, the detection and/or quantification algorithm
mentioned herein above and below might be a (deep) neural network,
a statistical approach known as a multivariate analysis of variance
(Manova), Support Vector Machines (SVM) or any other machine
learning algorithm, pattern recognition algorithm or statistical
approach. The detection algorithm might consist of three parts,
which are processed in a timely sequence, according to the three
drinking phases described above. This is, however, not necessary.
Instead of processing the multiple phases, e.g. two or three or
four phases or more, in a temporal manner and separately, the
drinking (and, by analogy, eating) procedure might also be detected
without distinguishing any phases. A deep learning algorithm might
find other criteria during a training phase.
[0043] For instance, instead of distinguishing the above-described
multiple phases, (e.g., two or three or any other number of phases)
of drinking and/or medication intake a priori in the step of
analyzing, a temporal dynamic behavior of the drinking process
and/or medication intaking process (in other words, the "history"
of the drinking process and/or medication intaking process), may be
incorporated by applying a Hidden Markow Model (HMM) or a recurrent
neural network (RNN), such as a long short-term memory (LSTM) or a
gated recurrent unit (GRU).
[0044] An artificial neural network or other machine learning
algorithms as described herein may be implemented directly in the
hearing device, e.g. in a chip with a high degree of parallel
processing, being integrated in the hearing device to this end.
Alternatively or additionally, this may also be implemented in any
other device of the hearing system. Since the analysis steps of
drinking or medication intake or eating detection and
quantification, or further items such as dehydration detection
described herein above and below, must not necessarily be solved in
real time and must not necessarily be all processed within the
hearing device itself, diverse and more complex algorithms and
processes than in the cited prior art are implementable.
[0045] In any of the embodiments described herein, the hearing
device or system may be configured to combine and analyse signals
of several sensors from the following exemplary list of sensors.
Corresponding detection criteria (as mentioned for each sensor)
are, in general, user-specific and/or use-case specific. This may
be learned by a suitable machine learning algorithm, as described
herein above and in the following. A combination of several sensors
may, for instance, be suitable to detect all the above-mentioned
phases of the drinking (or eating) procedure:
[0046] Accelerometers (or gyroscopes) in the hearing device may be
used to detect when the user elevates his head (e.g., slightly) or
tilts his head (e.g., within a certain range of tilting angles) to
take a sip. In contrast, to detect eating, the detection algorithm
may be based on the fact that the user is rather lowering his head
when taking a bite. As another example, to detect medication
intake, the detection algorithm may be based on the fact that the
user elevates his head even higher and/or tilts his head even more
to the back when taking medication.
[0047] In addition, the user might wear a smartwatch or a wristband
around his wrist or a finger ring at one of his fingers with
integrated movement sensors. While the user brings the glass to his
mouth and while drinking, his arm follows a specific gesture, which
may be recognized by these sensors so as to indicate the possible
action of drinking. This movement pattern might also be used to
distinguish between drinking, eating and other possible actions of
the user.
[0048] In addition, a drinking bottle might incorporate sensors and
be connected to the smart home (also known as smart water bottles).
The hearing device or system may be connected to the smart home as
well and receive those signals to incorporate them into the
detection process. Alternatively, those data may be collected in a
cloud or a smart home device to compute the algorithms.
[0049] Physiological sensors might be incorporated to detect intake
of liquid or medication or food: a physiological property of the
user as detected by such a sensor, e.g., a heart rate, a blood
pressure level, a blood analyte level, etc., can be influenced by
the drinking, eating and/or medication intake. E.g., the pulse of
the heart can be influenced and altered by generally swallowing
liquid or solid nutrition. This effect may also strongly vary
depending on the type and quantity of liquid or food taken by the
user, which might be incorporated in the analysis algorithm to
determine the respective values or detect the difference. Further,
certain types of medication, even medication unrelated to a
cardiovascular treatment, can have a secondary effect to alter a
blood pressure of the user after the medication intake, e.g. by
raising or lowering the blood pressure to a certain degree.
Detecting the blood pressure can thus be used to verify and/or
determine (qualitatively and/or quantitatively) an intake of
medication. Further, detecting a blood analyte level by a blood
analyte sensor (as described above) can give indications of a
consumed liquid and/or medication and/or nutrition.
[0050] Other sensors, such as electroencephalographic (EEG)
sensors, eye gaze or a camera within the smart home or integrated
into glasses might be incorporated for detection.
[0051] In addition, other sensors might be applied, such as:
[0052] Acoustic sensors: the microphones within the hearing device
or microphones in the ear canal might detect the gulping sound of a
liquid and/or the swallowing sound of medication or might improve
the detection rate by recognizing and/or excluding other actions,
such as own-voice or eating, and/or by recognizing different
actions of drinking and/or medication intake and/or eating allowing
to distinguish between those actions.
[0053] Voice Pickup sensors (VPUs) might be used to improve the
detection of gulping and/or swallowing of medication and
distinguish it from eating, and vice versa.
[0054] According to an embodiment, in the step of analyzing, a
dehydration risk of the hearing device user is estimated depending
on the determined values of the amount and/or of a frequency (i. e.
of how often he has drunk) of the user's liquid intake. The risk of
dehydration may be estimated depending on the drinking behavior,
e.g. the determined amount of liquid intake is compared with a
predetermined value "Dmin" which describes the minimum amount of
drinking the user needs to take. This value can be a fixed number
or might be calculated depending on the environmental conditions,
such as the ambient temperature, on the activity behavior of the
user, e.g. walking or sitting, and/or on medical conditions, such
as weight etc. The value "Dmin" can be adapted by the user via the
interactive user interface described below and/or by his doctor.
Determining a type of the ingested liquid, e.g., water as compared
to a different liquid type, as described above, may be employed to
further improve the estimation of a dehydration risk. For instance,
the amount of liquid intake may be only determined for an amount of
(pure) water intake and/or the determined amount of liquid intake
may be corrected with respect to a number of events in which (pure)
water has been ingested as compared to a number of events in which
a different liquid type has been ingested and/or a number of events
in which a dehydrating liquid type (e.g., coffee) has been
ingested. As another example, the amount of liquid intake may be
determined by weighting each drinking event with respect to the
type of the ingested liquid, wherein (pure) water is associated
with the largest weight. The weight can thus be indicative of a
contribution of a certain type of an ingested fluid to a hydration
of the user and/or an impact of the liquid type on the dehydration
risk. E.g., a liquid type containing sugar may be associated with a
smaller weight than (pure) water. A dehydrating liquid type may be
associated with an even smaller weight.
[0055] In this embodiment, the generated output may be configured
depending on the estimated dehydration risk so as to counsel the
user to ingest a lacking amount of liquid and/or so as to inform
the user and/or a person close to the user and/or a health care
professional about the estimated dehydration risk. Counselling the
user may be implemented e.g. using a smartphone app and/or
acoustically via an integrated loudspeaker of the hearing device
and/or via vibration on devices such as wristband, smartwatch or
finger ring, and/or via augmented reality using the user's glasses
to generate a virtual image of a glass or bottle.
[0056] Alternatively, instead of a dehydration system design where
users are admonished to drink more, the system design may enhance
the desire to drink when dehydration is detected. This type of
output depending on the estimated dehydration risk is described in
more detail further below.
[0057] According to an embodiment, an interactive user interface is
provided in the hearing system and the steps of analyzing and/or
generating an output are supplemented by an interaction with the
user via the interactive user interface, wherein the user is
enabled to input additional information pertaining to his liquid
and/or food and/or medication intake.
[0058] For example, such an interface may be configured such as to
enable the user to manually enter the data for drinking behaviour
and/or to correct and/or specify the liquid intake. It might
further provide the user with a possibility to add comments about
subjective descriptions and ratings about his health state, mental
condition, feelings, thirst, appetite, social condition (e.g. alone
or with company), cause of fluid intake, intake of medication, etc.
Thereby, for example, the hearing system might ask the user
proactively whether he has taken medication right after the system
has detected fluid intake. The proactive questioning might be
during a specific time predefined and stored in the hearing system,
such as the morning, the evening, just before or after taking a
meal, which, in turn, might be determined using the analysis
algorithms described herein. This proactive question shall help the
user to keep track about the intake of medication.
[0059] The interactive user interface might also help to improve
the accuracy of the detection in general or within a learning phase
of the analysis algorithm. It may also serve for a more elaborated
therapy.
[0060] The interactive user interface might comprise one or more of
the following types of user interface:
[0061] an app on a smartphone, smartwatch or another connected user
device, which provides a correction mode showing e.g. the detected
gulps right after detection.
[0062] An entry mode provided in the hearing device (e.g. by voice
recognition or via a button) or on a connected user device,
configured to enable the user to specify the liquid and/or food
and/or medication intake.
[0063] A conversational user interface incorporated in the hearing
device(s).
[0064] According to an embodiment, additional information about a
need to take a predetermined medication (e. g. on a regular basis)
is stored in the hearing system. In this embodiment, when a fluid
intake of the user is detected, the output is generated depending
on this additional information and comprises questioning the user,
via the interactive user interface, whether he has taken the
predetermined medication. The user's response to this question via
the interactive user interface may then be, for example, stored in
the hearing system and/or transmitted to the user and/or to a third
person close to the user and/or a health care professional, so as
to verify that the user has taken the predetermined medication.
[0065] Alternatively or additionally to questioning the user,
detecting the intake of medication may also be based on the fluid
intake detection in combination with the analysis of head movements
of the user, as described above. This may also be implemented by
the sensor-based machine learning algorithm as described herein
above and below.
[0066] According to an embodiment, in the step of generating an
output depending on the determined frequency and amount of the
liquid or, respectively, food or, respectively, medication ingested
by the user, an output configured such as to enhance the user's
desire to drink or, respectively, eat something, or, respectively,
take his medication is generated by augmented reality means in the
hearing system. In particular, this may be an output generated upon
estimating a dehydration risk in the embodiment described further
above, as alternative to an output which admonishes the user to
drink more. Such an output may also be selectively generated for
the enhancement of the drinking or medication intake or eating
desire, e.g. depending on a determined dehydration risk and/or
failure of medication intake and/or poor nutrition. For instance,
an output configured to enhance the user's desire to drink may only
be generated in situations in which the user shall be motivated for
the medication intake in conjunction with the drinking (e.g., in
situations in which the dehydration risk is low, but a risk
associated with a lack of medication intake has been determined as
rather high), but not during regular drinking activities.
[0067] In this embodiment, an output configured to enhance the
user's desire to drink or eat something, or to take (the
prescribed) medication, may, for example, be reached with suitable
sound, smell and video effects and methods. This may include
augmenting specific sounds or music associated by the user with
drinking or suitable to enhance a drinking experience. This may
also be achieved by presenting specific sounds or music associated
by the user with eating and/or which can lead to an enhanced
tasting-experience. The suitable effects may be very user-specific
and be determined as additional information by machine learning in
the analysis step of the present method, and/or indicated by the
user via the above-described interactive user interface.
[0068] The sounds or optical effects to be presented in order to
enhance the user's desire to drink, and/or to take medication,
e.g., during the drink, might, for example, strongly differ
dependent on the drink to be taken and on the user's cultural
background and music preference, which may be evaluated by the
hearing system in the analysis step. Same also applies to flavor
effects. Furthermore, it is also known that there may be a strong
user-specific correlation between smell and sound, for instance
bound to certain situations in life remembered by the user, so that
a suitable combination of smell and sound may lead to a still
stronger desire of the user to drink or, respectively, eat
something special in this embodiment. All this also applies to
video in addition to or instead of sound effects.
[0069] Specifically, one or more of the following outputs may be
generated by the hearing system in this embodiment:
[0070] An already existing drinking glass or bottle in the field of
vision of the user might be augmented visually with augmenting
colors representing the fluid, e.g. using smart glasses connected
to the hearing device. Similarly, a package or bottle containing
medication and/or a storage space used to store the medication
might be visually augmented,
[0071] the hearing system might display sounds, such as water
drops, sound which represents filling a glass with water or the
splashing sound of opening a beer, etc., which might augment the
desire to drink.
[0072] For elderly, where all senses decline, the enhancement of
smell might increase the experience as well. Depending on what the
user is drinking, the smart home connected to the hearing device
might pump smell into the room according to the drink. E.g. when
the content of the bottle or glass next to the user consists of
tea, the smell consists of a strong tea smell, when the glass
contains orange juice, the smell consists of fresh oranges.
[0073] Also a video might be played representing a drinking
experience and the corresponding sound. If the user is watching TV
(on a TV set connected to the hearing device) for a long time,
videos without the commercial content can be shown in between to
enhance the user's desire to drink or eat something or to take the
medication needed for the user's health.
[0074] According to an embodiment, when detecting that the user is
drinking or, respectively, eating something or, respectively,
taking medication and/or upon detecting which kind of liquid or,
respectively, food or, respectively, medication the user is taking,
an output configured such as to enhance the user's experience of
drinking or, respectively, eating or, respectively, taking
medication is generated by augmented reality means in the hearing
system depending on this (in particular multisensory) detection.
This shall be reached with similar methods as described for the
previous embodiment: with sound, smell and video. Furthermore, a
suitable output may be determined in a similar user-specific manner
as described above.
[0075] Furthermore, a similar algorithm as described herein can be
applied to monitor any other health related activities and body
symptoms, such as e.g. mentioned at the outset.
[0076] The above-described embodiments, where the output is
generated such as to enhance the user's desire to ingest some
healthy liquid or food or prescribed medication or to enhance the
respective eating or drinking or medication intake experience,
might be particularly applied to the elder users.
[0077] Thereby, for example, for the elderly who have lost their
appetite, the tasting experience may be enriched with multisensory
cues and sounds designed such as to enhance the eating or drinking
experience. Chewing sounds may be augmented so as to let the meal
appear more fresh and/or crisp. In particular, by a suitable sound
stimulus presented to the hearing device user, the perceived taste
of food may be influenced (e.g. low pitch sounds relate to bitter,
salty flavors. High pitch sounds relate to sweet, sour flavors).
This may be determined so as to stimulate the consumption of
certain types of aliments, which better suit the dietary
requirements of the user.
[0078] To support the determination of a suitable output and/or to
support the sensor-based analysis algorithm, the user may enter the
kind of drink or food or medication via the above-described user
interface, or a barcode on the drink or food or medication package
may be scanned with the connected user device such as smartphone or
a smartwatch.
[0079] By the method described herein, for example, obesity and a
risk of heart diseases due to bad eating (or also drinking) habits
of the users may be effectively reduced. Further, a health risk
associated with an irregular or neglected or omitted medication
intake can be effectively reduced.
[0080] Further aspects relate to a computer program for detecting
and quantifying a liquid and/or food and/or medication intake of a
user wearing a hearing device which comprises at least one
microphone, which program, when being executed by a processor, is
adapted to carry out the steps of the method as described above and
in the following as well as to a computer-readable medium, in which
such a computer program is stored.
[0081] For example, the computer program may be executed in a
processor of a hearing device, which hearing device, for example,
may be carried by the person behind the ear. The computer-readable
medium may be a memory of this hearing device. The computer program
also may be executed by a processor of a connected user device,
such as a smartphone or smartwatch or any other type of mobile
device, which may be a part of the hearing system, and the
computer-readable medium may be a memory of the connected user
device. It also may be that some steps of the method are performed
by the hearing device and other steps of the method are performed
by the connected user device.
[0082] In general, a computer-readable medium may be a floppy disk,
a hard disk, an USB (Universal Serial Bus) storage device, a RAM
(Random Access Memory), a ROM (Read Only Memory), an EPROM
(Erasable Programmable Read Only Memory) or a FLASH memory. A
computer-readable medium may also be a data communication network,
e.g. the Internet, which allows downloading a program code. The
computer-readable medium may be a non-transitory or transitory
medium.
[0083] Further aspects relate to a hearing device worn by a hearing
device user and to a hearing system comprising such a hearing
device, as described herein above and below. The hearing device or,
respectively, the hearing system is adapted for performing the
method described herein above and below. As already mentioned
further above, the hearing system may further include, by way of
example, a second hearing device worn by the same user and/or a
connected user device, such as a smartphone or smartwatch or other
mobile device, or personal computer, used by the same user.
[0084] According to an embodiment, the hearing device comprises: a
microphone; a processor for processing a signal from the
microphone; a sound output device for outputting the processed
signal to an ear of the hearing device user; and, as the case may
be, a transceiver for exchanging data with the connected user
device and/or with another hearing device worn by the same
user.
[0085] It has to be understood that features of the method as
described above and in the following may be features of the
computer program, the computer-readable medium and the hearing
device and the hearing system as described above and in the
following, and vice versa.
[0086] These and other aspects will be apparent from and elucidated
with reference to the embodiments described hereinafter.
[0087] FIG. 1 schematically shows a hearing system 10 including a
hearing device 12 in the form of a behind-the-ear device carried by
a hearing device user (not shown) and a connected user device 14,
such as a smartphone or a tablet computer or any other body worn
device, e.g., a smart watch, a wrist band, or a finger ring. It has
to be noted that the hearing device 12 is a specific embodiment and
that the method described herein also may be performed with other
types of hearing devices, such as in-the-ear devices.
[0088] The hearing device 12 comprises a part 15 behind the ear and
a part 16 to be put in the ear canal of the user. The part 15 and
the part 16 are connected by a tube 18. In the part 15, at least
one microphone 20, a sound processor 22 and a sound output device
24, such as a loudspeaker, are provided. The microphone(s) 20 may
acquire environmental sound of the user and may generate a sound
signal, the sound processor 22 may amplify the sound signal and the
sound output device 24 may generate sound that is guided through
the tube 18 and the in-the-ear part 16 into the ear canal of the
user.
[0089] The hearing device 12 may comprise a processor 26 which is
adapted for adjusting parameters of the sound processor 22 such
that an output volume of the sound signal is adjusted based on an
input volume. These parameters may be determined by a computer
program run in the processor 26. For example, with a knob 28 of the
hearing device 12, a user may select a modifier (such as bass,
treble, noise suppression, dynamic volume, etc.) and levels and/or
values of these modifiers may be selected, from this modifier, an
adjustment command may be created and processed as described above
and below. In particular, processing parameters may be determined
based on the adjustment command and based on this, for example, the
frequency dependent gain and the dynamic volume of the sound
processor 22 may be changed. All these functions may be implemented
as computer programs stored in a memory 30 of the hearing device
12, which computer programs may be executed by the processor 26
and/or sound processor 22.
[0090] The hearing device 12 further comprises a transceiver 32
which may be adapted for wireless data communication with a
transceiver 34 of the connected user device 14, which may be a
smartphone or tablet computer. It is also possible that the
above-mentioned modifiers and their levels and/or values are
adjusted with the connected user device 14 and/or that the
adjustment command is generated with the connected user device 14.
This may be performed with a computer program run in a processor 36
of the connected user device 14 and stored in a memory 38 of the
connected user device 14. The computer program may provide a
graphical user interface 40 on a display 42 of the connected user
device 14.
[0091] For example, for adjusting the modifier, such as volume, the
graphical user interface 40 may comprise a control element 44, such
as a slider. When the user adjusts the slider, an adjustment
command may be generated, which will change the sound processing of
the hearing device 12 as described above and below. Alternatively
or additionally, the user may adjust the modifier with the hearing
device 12 itself, for example via the knob 28.
[0092] The user interface 40 also may comprise an indicator element
46, which, for example, displays a currently determined listening
situation.
[0093] The hearing device 12 further comprises at least one further
sensor 50, the position of which in the part 15 of the hearing
device 12 is to be understood as only symbolical, i.e. the further
sensors 50 may also be provided in other parts and at other
positions in the hearing device 12. The microphone(s) 20 and the
further sensor(s) 50 enable the hearing device 12 or the hearing
system 10 to perform a multi-sensor-based analysis for detecting
and quantifying a liquid and/or food intake of a user as described
herein. The further sensor(s) 50 may be integrated not only in the
hearing device 12, but additionally also in the connected user
devices such as the connected user device 14 shown in FIG. 1, or a
wearable such as a wrist band or smart watch or finger ring, etc.
(not shown). As described in more detail herein above and below,
the further sensors 50 may include any types of physical or
physiological sensors--e.g. movement sensors, such as
accelerometers, and/or optical sensors, such as cameras, and/or
temperature sensors and/or heart rate sensors etc.
[0094] The hearing system 10 shown in FIG. 1 can be adapted for
performing a method for detecting and quantifying a liquid and/or
food intake of a user wearing the hearing device 12, as described
in more detail herein above and below.
[0095] In some implementations, the hearing system 10 comprises the
hearing device 12 without the connected user device 14. E.g., the
hearing system 10 may consist of the hearing device 12, or the
hearing system 10 may consist of two hearing devices 12 configured
two be worn at a respective ear of the user in a binaural
configuration. Such a hearing system can also be adapted for
performing a method for detecting and quantifying a liquid and/or
food intake of a user wearing the hearing device 12 (or two hearing
devices), as described in more detail above and below.
[0096] FIG. 2 shows an example for a flow diagram of this method
according to an embodiment. The method may be a
computer-implemented method performed automatically in the hearing
system 10 of FIG. 1.
[0097] In a first step S10 of the method, an audio signal from the
at least one microphone 20 and/or a sensor signal from the at least
one further sensor 50 is received, e.g. by the sound processor 22
and the processor 26 of the hearing device 12.
[0098] In a second step S20 of the method, the signal(s) received
in step S10 are collected and analyzed in the hearing device 12 or
in the hearing system 10, so as to detect each time the user drinks
and/or eats something and/or takes medication, wherein drinking
and/or medication intake is distinguished from eating and/or
wherein drinking is distinguished from medication intake, and so as
to determine values indicative of how often this is detected and/or
a respective amount of liquid and/or food and/or medication
ingested by the user. The step of analyzing may include applying
one or more machine learning algorithms, which may be implemented
on a processor chip integrated in the hearing device 12, or
alternatively in the connected user device 14 or somewhere else in
the hearing system or in a remote server or cloud connected to
it.
[0099] In a third step S30 of the method, the determined values are
stored in the hearing system 10 and, based on the stored values, a
predetermined type of output as described in more detail herein
above and below is generated, for example, to the user and/or to a
third party, such as a person close to the user and/or a health
care professional.
[0100] In the following, an illustrative method is illustrated, by
way of example only, for estimating a risk of dehydration of the
user and generating a corresponding output to counsel the user to
drink more, and for verifying a medication intake based on the
detected liquid intake. It may also be applied in a similar manner
for other use cases described in detail herein, such as enhancing
the user's desire to drink and/or to take medication or enhancing
the eating or drinking or medication taking experience of the
user.
[0101] In this example, the solution according to the present
method comprises a so-called "sensory part", which may cover the
above step S10 and at least a part of step S20 and comprise
detecting how many times and how long the user is drinking. It
further comprises a so-called "actuator part", which may cover the
above step S30 and at least a part of step S20 and comprise storing
the detected values into the memory of the hearing device or into
the memory of an additional device such as a smartwatch, a
wristband, a finger ring, a belt, a cloud etc.; estimating the risk
of dehydration and counselling the user and/or, if the user agrees,
informing a third party about his drinking behavior, such as a
doctor, nurses or caregivers. That is, if the user agrees, his data
and values determined in the analysis step of the method are sent
to third parties (in addition or alternatively to counseling the
user). The sensory part is capable to detect how often and how many
swallows or gulps the user takes each time. It is also capable to
distinguish between drinking and eating, own-voice and other
possible confusions. Three (to four) phases of drinking might be
distinguished, such as:
[0102] Phase 1: Grabbing a glass or bottle and bring it to the
mouth
[0103] Phase 2: Gulping the liquid
[0104] Phase 3: Putting the glass back on the table. This may also
include detecting the sub-phases of (3a) a movement to bring the
glass/bottle from the mouth down to the table and (3b) the actual
putting of the glass on the table.
[0105] Different sensors might be better suited to detect the
different phases, for example:
[0106] Phase 1: movement and/or orientation sensors, e.g. within a
wrist band or a smart watch or a finger ring and/or in a hearing
device, might be most suited,
[0107] Phase 2: microphones within the hearing device and
accelerometers within the ear canal might be most suited,
[0108] Phase 3: movement and/or orientation sensors, e.g. on a
wrist band or smart watch or finger ring, and microphones might be
most suited.
[0109] The hearing device 12 or system 10 may combine and analyze
signals of several sensors to detect all phases of the drinking
procedure:
[0110] Accelerometers (or gyroscopes) in the hearing device may be
used to detect when the user elevates his head slightly to take a
sip. In contrast, when eating, the user is rather lowering his head
when taking a bite.
[0111] In addition, the user might wear a smartwatch or a wristband
around his wrist or a ring around one of his fingers with
integrated movement sensors. While the user brings the glass to his
mouth and while drinking, his arm follows a specific gesture, which
indicates the possible action of drinking. This movement pattern
might also be used to distinguish between drinking and other
possible actions of the user.
[0112] In addition, a drinking bottle might incorporate sensors and
be connected to the smart home (also known as smart water bottles).
The hearing device 12 or system 10, also connected to the smart
home, receives those signals to incorporate them into the detection
process. Alternatively, those data are collected in a cloud or a
smart home device to compute the algorithms.
[0113] Physiological sensors, e.g., heart rate sensors and/or blood
pressure sensors and/or temperature sensors, might be incorporated
to detect fluid intake: the pulse of the heart and/or blood
pressure and/or body temperature can be influenced and altered by
swallowing liquid or solid nutrition. Further, the physiological
sensors may by employed to determine a type of the ingested fluid,
e.g. a blood pressure may be altered to a larger extent by
consuming coffee as compared to consuming pure water.
[0114] Other sensors, such as electroencephalographic (EEG)
sensors, eye gaze or a camera within the smart home or integrated
into glasses might be incorporated for detection
[0115] Physiological sensors, e.g., a sensor sensitive to an
analyte in a body fluid such as blood or sweat, might be
incorporated to detect a physiological effect of the fluid intake:
the quantity of water contained in blood or tissue can be altered
by the fluid intake. Further, the physiological sensors may by
employed to determine a type of the ingested fluid, e.g.
determining a glucose level can indicate whether the ingested fluid
is pure water or a sugary drink and/or determining a caffeine level
and/or alcohol level can indicate whether the ingested fluid is a
dehydrating fluid (e.g., coffee or an alcoholic beverage) rather
than a hydrating fluid (e.g., water). Determining the type of the
ingested fluid can be employed to render the estimation of a
dehydration risk of the user more exact, e.g., by associating each
type of the ingested fluid with a specific weight indicative of its
contribution to a hydration of the user, as described above.
[0116] In addition, other sensors might be applied, such as:
[0117] Acoustic sensors: the microphones within the hearing device
or microphones in the ear canal might detect the gulping sound or
might improve the detection rate by excluding other actions, such
as own-voice or eating.
[0118] Voice Pickup sensors (VPUs) might be used to improve the
detection of gulping and distinguish it from eating.
[0119] The outcome of the several sensors is fed into a detection
algorithm. The detection algorithm might consist of three parts,
which are processed in a timely sequence, according to the three
drinking phases described above. The algorithms might be a (deep)
neural network, multivariate analysis of variance (Manova), Support
Vector Machines (SVM) or any other machine learning algorithm,
pattern recognition algorithm or statistical approach.
[0120] Instead of processing the three phases in a temporal manner
and separately, the drinking procedure might be also detected
without distinguishing any phases. A deep learning algorithm might
find other criteria during a training phase. Since dehydration and
drinking detection must not be solved in real time and must not be
processed within the hearing device 12, diverse and more complex
algorithms and processes are applicable.
[0121] Description of the method based on (deep) neural
networks:
[0122] Input: data sets x=(t, x1, x2, x3, . . . ) for each time
instance (t), where x1, x2, x3, . . . represent the collected
sensor values, as described above.
[0123] Output data: y=(t, fi, d) for the input time instance (t),
where (fi) is the determined number and (d) the determined duration
of fluid intake in this example.
[0124] Supervised learning: the algorithm is trained with labeled
data.
[0125] The learning part can be supervised and trained with a
database of sensor data. Here, a suitable database of training data
based on the audio signal from a microphone contains recorded
swallowing or gulping sound data of a large number of people--in
this example 1000 randomly selected persons. Alternatively or
additionally, representative sound data of drinking, swallowing,
gulping, chewing etc. available in the Internet may be used as
training data. The training can also be unsupervised by learning in
an environment with more information available, such as using smart
water bottles and/or information within a smart home and/or input
from the user. Reinforcement learning or "deep reinforcement
learning", as a third category of machine learning paradigms beside
supervised and unsupervised learning, proposes different
algorithms, which may be applied in the present method as well.
[0126] Instead of distinguishing three phases of drinking a priori,
the "history" of the drinking process or, respectively, its
temporal dynamic behaviour, might be incorporated by applying
different machine learning methods. Beside the traditional method
of Hidden Markow Model (HMM), recurrent neural networks (RNN) might
be applied, e.g.
[0127] Long short-term memory (LSTM)
[0128] Gated recurrent unit (GRU).
[0129] GRU might be more suited to the application due to less
computational effort.
[0130] Furthermore, the analysis may be implemented by parallel
decision units in the hearing device 12 or in the hearing system
10. That is, instead of using one algorithm, several machine
learning (ML) algorithms might be applied in parallel to compute
the detection probabilities.
[0131] This may, for example, allow for the following features:
[0132] To detect some drinking phases, one or the other sensor
and/or algorithm might be more suitable than the other.
[0133] Weighting parameters (e.g. in the activation functions of
the artificial neurons) might be separately introduced and
determined for each drinking phase and each decision algorithm.
[0134] An algorithm in sequence to the several decision algorithms,
e.g. in an architecture such as a decision tree, might be used to
make the final decision.
[0135] As already mentioned, the sensor data might be collected and
analyzed in different types of analysis units, for example:
[0136] within the hearing device 12, the hearing system 10
incorporating additional devices such as smartphone 14, smartwatch,
wrist band, remote control, remote microphone, hearing aid/device
etui, TV-Connector, . . .
[0137] on a server, cloud, smart home device.
[0138] As already mentioned above, following on the above-described
"sensor part", the "actuator part" of the method consists of
several steps in this example: storing the detected values in a
memory, estimation of a dehydration risk, counselling the user,
informing third party about drinking behavior of the user. Two of
these steps are further described in the following:
[0139] The risk of dehydration is estimated dependent on the
detected drinking behavior, e.g. the determined amount of liquid
ingested by the user is compared with a certain value "Dmin"
describing the necessary minimum amount of drinking. This value can
be a fixed number or might be calculated depending on one or more
of the following factors:
[0140] on the environmental condition, such as the weather
temperature: if the temperature around the user is very high, the
value "Dmin" is increased;
[0141] on the activity behavior of the user: if he walks a lot
(e.g. as detected by an accelerometer) or performs an activity with
high physical effort (the heart beat is increased, e.g. as measured
by a corresponding sensor), the value "Dmin" increases
[0142] on medical conditions (e.g. as preset by the user's
physician)
[0143] The values "Dmin" might be stored in a table including
predetermined values or calculated with a regression equation, e.g.
as known from medical studies. Alternatively or in addition, the
value "Dmin" can also be adapted by the user or a doctor.
[0144] The above-mentioned step of counseling the user might
include several options, such as one or more of the following:
[0145] a smartphone app on the connected user device 14 displays
how much the user has drunk and how much the user should have drunk
during the day.
[0146] the user gets notified when he shall drink: acoustically,
via vibration on devices such as wristband, smartwatch, or via
augmented reality. This may, for example, be implemented in that
the user's glasses (or smart glasses being a part of the hearing
system 10 or connected to it) augment visually a drinking glass or
a drinking bottle.
[0147] Moreover, in the present example, an interactive user
interface is provided in the hearing system 10 and the steps of
analyzing (S20) and/or generating an output (S30) are supplemented
by an interaction with the user via this user interface, wherein
the user is enabled to input additional information pertaining to
his liquid intake.
[0148] For example, such an interface may be configured such as to
enable the user to manually enter the data for drinking behavior
and/or to correct and/or specify the liquid intake. It might
further provide the user with a possibility to add comments about
subjective descriptions and ratings about his health state, mental
condition, feelings, thirst, appetite, social condition (e.g. alone
or with company), cause of fluid intake, intake of medication,
etc.
[0149] Thereby, for example, the hearing system 10 might ask the
user proactively whether he has taken medication right after the
system has detected fluid intake in step S20. The proactive
questioning might be during a specific time predefined and stored
in the hearing system 10, such as in the morning, in the evening,
just before or after taking a meal, which, in turn, might be
determined using the analysis algorithm in step S20. This proactive
question shall help the user to keep track about the intake of
medication.
[0150] The interactive user interface might also help to improve
the accuracy of the detection in general or within a learning phase
of the analysis algorithm in step S20. It may also serve for a more
elaborated therapy.
[0151] The interactive user interface (also denoted as UI) might
comprise one or more of the following types of user interface:
[0152] An app on a smartphone (connected user device 14 in FIG. 1),
smartwatch or another connected user device, which provides a
correction mode showing e.g. the detected gulps right after
detection (e.g. the graphical user interface 40 on the display 42
of the connected user device 14).
[0153] An entry mode provided in the hearing device (e.g. via the
slider 44 in the graphical user interface 40 on the connected user
device 14), configured to enable the user to specify the liquid
intake.
[0154] A conversational user interface incorporated in the hearing
device 12.
[0155] In some implementations of the above described method,
drinking may not be distinguished from a medication intake, i.e.,
it may not be separately determined whether an action of drinking
is related to the drinking of a fluid (such as water) without any
accompanying medication intake (such as swallowing a pill) and
another action of drinking is carried out for the purpose of a
medication intake (e.g., swallowing water to facilitate a
swallowing of a pill). Those implementations may be beneficial to
provide an estimation of the dehydration risk with a lower effort,
e.g. under the assumption that a medication intake will not alter
the dehydration risk to a considerable amount. In some other
implementations of this method, drinking can be further
distinguished from a medication intake, e.g., it may be determined
whether the drinking is accompanied by a medication intake or not.
Those implementations may be beneficial to provide an estimation of
the dehydration risk with an increased accuracy, e.g. in certain
cases of a medication which can impact the dehydration risk to a
considerable amount. Further, the proactive inquiry to the user by
the hearing system 10 whether he has taken medication right after
the system has detected fluid intake in step S20 may be carried out
depending whether the medication intake has been detected or not,
e.g., to remind the user only in those cases to take his medication
in which it seems appropriate due to a lack of a detected
medication intake.
[0156] In the following, the method will be further illustrated, by
way of another example, for estimating a risk of an irregular or
omitted medication intake of the user and generating a
corresponding output to counsel the user to take his medication
more regularly. It may also be applied in a similar manner for
other use cases described in detail herein, such as enhancing the
user's desire to drink and/or to take medication or enhancing the
eating or drinking or medication taking experience of the user.
[0157] The method of estimating a risk of an irregular and/or
omitted medication intake may comprise any of the steps S10, S20,
and S30 described above in conjunction with estimating a risk of
dehydration of the user, wherein, however, drinking is
distinguished from medication intake, and those steps are modified
according to the following description. During the gulping of the
liquid (phase 2), a tilting of the user's head may be determined,
e.g., be determining a tilting angle of the head relative to the
surface of the earth by a movement sensor and/or an orientation
sensor. Drinking (without medication intake) may then be
distinguished from a medication intake (which may be accompanied by
drinking) by determining whether the tilting angle is in a specific
angular range associated with the medication intake (e.g.,
swallowing of a pill) rather than in another angular range
typically occurring during a (regular) drinking not involving a
medication intake.
[0158] Further, before the grabbing a glass or bottle and bringing
it to the mouth (phase 1), another phase associated with the
medication intake may be determined, e.g., grabbing the medication
(e.g., grabbing a pill and/or opening a pill dispenser and taking a
pill) and bringing it to the mouth. This phase (which may be
denoted as "phase 0") may be detected correspondingly to "phase 1"
by movement sensors and/or orientation sensors, e.g. in a body worn
device (such as wrist band, smartwatch, finger ring, belt, etc.)
and/or included in a hearing device, as described above. E.g.,
"phase 0" can be distinguished from "phase 1" by a different
movement type and/or movement pattern. In this way, in cases in
which such a "phase 0" occurring before the "phase 1" is
determined, a medication intake (which may be accompanied by
drinking) can also be distinguished from drinking without
medication intake, i.e. from cases in which "phase 1" may be
detected without a preceding "phase 0".
[0159] Further, after the medication and/or liquid has been
ingested (i.e., after "phase 2" or after "phase 3"), an impact on a
physiological property of the user caused by the medication intake
and/or drinking may be determined, e.g. a change of a heart rate
and/or change of a blood pressure and/or a change a blood analyte
level. Depending on an occurrence of such a change of the
physiological property and/or depending on an extent to which a
change of the physiological property occurs, it may be determined
whether the medication has been taken and/or drinking (without the
medication intake) may thus be distinguished from a medication
intake (which may be accompanied by drinking). To illustrate, a
change of blood pressure may be rather identified as a side effect
of a medication intake as compared to a drinking of (pure) water
and may therefore serve as an indicator of the medication intake. A
change of blood pressure, however, may also be caused by certain
beverages (e.g., coffee). To distinguish the medication intake from
the drinking of such beverages (without medication intake), a
certain extent of the blood pressure change may be determined which
may then be associated with the medication rather than the drinking
of such beverages and/or, in a case in which a decrease of the
blood pressure is determined, it may be directly employed as an
indicator of the medication intake (e.g., because drinking of those
beverages typically only leads to a blood pressure increase,
whereas certain types of medication can also cause a blood pressure
decrease). Furthermore, additional indicators of the medication
intake may be determined (e.g., a presence or absence of an analyte
indicative of certain drug component in a body fluid) to increase
the medication intake determination accuracy. Detecting this phase
of an impact on a physiological property of the user caused by the
medication intake and/or drinking (which may be denoted as "phase
4") may be performed by one or more physiological sensors
configured to provide physiological signals indicative of a
physiological property of the user, e.g. cardiovascular sensors,
heart rate sensors, blood pressure sensors, blood analyte sensors,
body fluid analyte sensors, etc., as described above. Those
physiological sensors may be included in the hearing device and/or
another body worn device (e.g., a smartwatch, wristband, finger
ring, belt, etc.).
[0160] In order to detect any of the different phases described
above, a suitable detection algorithm can be employed analogous to
the detection algorithms described above in conjunction with the
estimation of a dehydration risk (for instance, by a separate
processing of each of the multiple phases, e.g., in a temporal
manner, or by a processing without separating the different phases
and/or by a separate processing of some phases and non-separate
processing of some other phases).
[0161] A risk of an irregular and/or omitted medication intake may
then be estimated dependent on the determined medication intake
behavior. E.g., the determined amount of medication taken by the
user may be compared with a certain value "Mmin" describing the
necessary minimum amount of medication, which may be entered by the
user or a medical professional and might be stored in a table,
e.g., in a memory of the hearing device or another user device.
Further, a risk of an excessive medication intake may also be
estimated dependent on the determined medication intake behavior.
E.g., the determined amount of medication taken by the user may be
compared with a certain value "Mmax" describing a maximum amount of
medication which should not be exceeded, e.g., to avoid negative
medical side effects or an over-medication. The value "Mmax" which
may also be entered by the user or a medical professional and
stored in a memory. Counseling the user with regard to the
medication intake might include several options, as also described
above in conjunction with the dehydration risk, e.g., via
displaying a required medication intake versus the actual
medication intake on a user device 14 and/or by providing user
notifications to remind the user about the medication intake which
may occur, e.g., during or after a detected drinking activity of
the user and/or when a risk associated with an irregular and/or
omitted and/or excessive medication intake has been determined.
[0162] While the invention has been illustrated and described in
detail in the drawings and foregoing description, such illustration
and description are to be considered illustrative or exemplary and
not restrictive; the invention is not limited to the disclosed
embodiments.
[0163] Other variations to the disclosed embodiments can be
understood and effected by those skilled in the art and practicing
the claimed invention, from a study of the drawings, the
disclosure, and the appended claims. In the claims, the word
"comprising" does not exclude other elements or steps, and the
indefinite article "a" or "an" does not exclude a plurality. A
single processor or controller or other unit may fulfill the
functions of several items recited in the claims. The mere fact
that certain measures are recited in mutually different dependent
claims does not indicate that a combination of these measures
cannot be used to advantage. Any reference signs in the claims
should not be construed as limiting the scope.
LIST OF REFERENCE SYMBOLS
[0164] 10 hearing system [0165] 12 hearing device [0166] 14
connected user device [0167] 15 part behind the ear [0168] 16 part
in the ear [0169] 18 tube [0170] 20 microphone(s) [0171] 22 sound
processor [0172] 24 sound output device [0173] 26 processor [0174]
28 knob [0175] 30 memory [0176] 32 transceiver [0177] 34
transceiver [0178] 36 processor [0179] 38 memory [0180] 40
graphical user interface, interactive user interface [0181] 42
display [0182] 44 control element, slider [0183] 46 indicator
element [0184] 50 further sensor, in particular a movement and/or
orientation sensor, e.g., an accelerometer, and/or a physiological
sensor, e.g., a photoplethysmography (PPG) sensor and/or an
electrocardiography (ECG) sensor and/or a blood analyte sensor
* * * * *